Skip to main content
← Back to Docs
⟁EV Scanner

Market Heatmap

Spot category-wide moves at a glance with the heatmap

The Market Heatmap (/dashboard/heatmap) is your panoramic, single-screen view of where the model currently sees the most mispricing on Polymarket (Kalshi markets do not feed this page). Instead of scrolling a flat list of markets, the active board is collapsed into the top eight categories by volume, each one a tile. A tile's area is proportional to the category's traded volume, and its colour tracks the category's average AI edge — so size tells you where the money is and colour tells you where the model disagrees with it. Under the number you get the market count and the category's volume as plain text. In one glance you can tell which categories the model considers efficiently priced and which it thinks are out of line.

This page is built for situational awareness, not for execution. It answers "what is moving and where should I look?" so that by the time you open the EV Scanner or a specific market, you already know which corner of the board deserves your attention.

What the Heatmap shows

The Heatmap analyzes the 100 most active markets on each refresh and aggregates them into tiles. There are two layers stacked on the page:

  • •Category tiles (top grid). The top eight categories by volume — Politics, Crypto, Sports, Economics, Pop Culture, and so on. Each tile is sized relative to its share of total volume and colored by how much the markets inside it are moving. The big number on each tile is the category's average edge in percentage points (pp) — the average gap between Predite's AI-modeled fair probability and the current market price across every market in that category. Under it you get the market count and total volume (e.g. 42 markets · $3.1M).
  • •Top Markets (lower panel). The six individual markets with the largest absolute edge right now, regardless of category. Each row shows the question, its category and volume, a one-line snippet of the AI's reasoning, and the signed edge (+12.4pp in teal for a YES lean, -9.1pp in rose for a NO lean). This is the bridge from "this whole sector is moving" to "here is the specific contract behind it."

Together, the two layers let you move from macro to micro without leaving the page: read the grid to find the hot category, then drop your eyes to the Top Markets panel to see which contracts are driving it.

How tile size and color work

Two independent visual channels carry the information, and reading them together is the whole skill:

  • •Colour = average AI edge for the category. A pale tile is a category the model thinks is fairly priced; a saturated one is a category where it sees a wide gap to fair value. Colour is driven by the model's opinion, not by how much prices moved — a category can be violently volatile and still show pale if the model agrees with every new price.
  • •Size = volume. A tile's area is proportional to the category's traded volume, so the board tells you where the money is before you read a single number. The volume figure is still printed under the edge — use it when two tiles look close. This is also the fastest way to spot the trap: a bright colour on a small tile is a large modeled edge in a category one order can move.
  • •The tile number has no sign. It averages the *absolute* edge of the markets in the category, so it is a measure of magnitude only. A category where the model is bullish on half its markets and bearish on the other half still shows a large positive number. For direction you have to drop to the Top Markets rows, which do show the signed edge.

The combination is what matters. The tile you want is large and saturated: a deep, liquid category that is also repricing fast. That is where a real edge and the ability to act on it overlap.

How to read it, step by step

  1. Scan for the brightest tiles first. Ignore the pale ones — they are categories where the model has nothing to say. Then check the volume line: a bright tile over a thin category is the one to distrust.
  2. Check the headline pp number. A category averaging +8.0pp or more means the AI consistently sees markets there as mispriced. A category hovering near 1–2pp is efficiently priced — the crowd and the model agree, and there is little to do.
  3. Compare the tile to its neighbors. One hot tile in an otherwise calm board is usually a single-event story (a poll dropped, a token pumped). Several adjacent tiles lighting up at once is a regime shift — a macro print, a debate, a geopolitical headline repricing an entire theme. Regime shifts are the highest-value thing this page catches.
  4. Drop to the Top Markets panel. Find the rows whose category matches your hot tile. The signed edge and the reasoning snippet tell you the direction and the *why* in one line.
  5. Hand it to the Scanner. Tiles and Top Markets rows are not links — the only clickable control on a row is its trade button. To dig in, switch to the EV Scanner and filter to that category for the full ranked list of edges, the sizing math and the AI rationale.

Spotting regime shifts and category-wide moves

This is the Heatmap's signature use. A regime shift is when a whole category repriced because the world changed, not because one contract wandered. You recognize it on the grid by clustering: instead of one bright tile, you see a block of related tiles intensify together between refreshes. Politics and Economics lighting up simultaneously around a Fed decision; Crypto and a related macro tile moving on the same headline; multiple Sports sub-markets shifting as injury news breaks.

Why it pays to catch this early: when a category moves as a block, individual markets inside it often lag the repricing by minutes. The leaders move first, the stragglers catch up. The Heatmap shows you the block is moving; the Top Markets panel and the Scanner show you which specific contracts haven't caught up yet. That lag is the tradable window.

A practical habit: glance at the Heatmap at the start of each session and around scheduled events (see the News & Calendar feature). If a category is brighter than when you last looked, something happened while you were away — start your research there instead of from the top of an alphabetical list.

Filtering by category

The default view shows the top eight categories by volume, which is the right starting point for a board-wide scan. To go deeper into one theme:

  • •Take the category to the Scanner. The Heatmap has no filters or controls of its own — it is a read-only board. Once a tile catches your eye, open the EV Scanner and use its category dropdown to get the ranked, market-by-market list behind that tile.
  • •Compare categories by toggling between them to judge *relative* heat. A 5pp average might be hot for normally-efficient Economics markets but cool for a volatile Crypto board. Context matters; the same number means different things in different categories.
  • •Narrow to your edge. If you only trade Politics or only trade Sports, filter to it and treat the Heatmap as a focused dashboard for your niche rather than a firehose.

When you find a category worth tracking, set a rule on the Alerts page (/dashboard/alerts) so you are pinged without keeping the Heatmap open all day. Match the rule to what alerts can actually express: you pick a scope — all markets, or Politics, Crypto, Sports or Finance — and then any combination of a minimum edge in pp, a minimum price move in pp, and a single market by id. Conditions stack: each one you fill has to pass. The move is measured against the last signal for that market before the current 30-minute window, so a market seen for the first time never fires a move rule.

How it complements the EV Scanner

The Heatmap and the EV Scanner run on the same underlying AI probability engine — the same multi-model consensus (Claude analysis blended with heuristic and microstructure models) that produces the edge figure on every market. The difference is altitude and intent:

  • •Heatmap = discovery and breadth. Aggregated, visual, fast. It tells you *where* to look across the whole board and catches category-wide moves a flat list hides. It is the radar.
  • •EV Scanner = precision and depth. A ranked, filterable, sortable list of individual markets with full edge, confidence, suggested direction (YES/NO/HOLD), and complete AI reasoning. It tells you *exactly what* to trade and at what size. It is the targeting computer.

The intended workflow is Heatmap → Scanner → trade:

  1. Heatmap to spot the hot category or the regime shift.
  2. Scanner (filtered to that category) to rank the specific markets, read full reasoning, and confirm the edge survives scrutiny.
  3. Kelly Calculator to size the position, then paper-trade or live-trade it.

Using the Scanner without the Heatmap means you can miss a category-wide move because you were filtered into the wrong sector. Using the Heatmap without the Scanner means you see heat but lack the per-market detail to act safely. They are designed as a pair.

Tips

  • •Big and bright beats either alone. A saturated tiny tile is usually one whale or a stale quote. A pale huge tile is a liquid market the crowd has already priced correctly. The opportunity lives where size and color overlap.
  • •Direction comes from the number, not the color. Color is magnitude only. Always read the signed pp value before assuming which side of the market is the play.
  • •Treat clusters as the headline. One hot tile is a story; a cluster of hot tiles is a *regime*. Regimes are where the lagging-market opportunities hide.
  • •Re-check after events. The board after a Fed print, a debate, or a major sports result looks completely different. Make a post-event Heatmap glance part of your routine.
  • •Confirm before sizing. The Heatmap is a pointer. Always confirm the specific edge in the Scanner and read the full AI reasoning before committing capital.

Gotchas

  • •It refreshes about every two minutes, not tick-by-tick. The Heatmap is cached and revalidates on a roughly 120-second cycle to stay fast and keep API load down. The Realtime Stream is fresher — it re-polls every 30 seconds — but it is still polling, not a trade-by-trade feed. Nothing in Predite streams individual fills. Do not treat a Heatmap tile as a live quote.
  • •Thin categories distort easily. A category with only a handful of markets can show a dramatic average edge off one outlier. Always sanity-check the market count and volume under the headline number before trusting it.
  • •Color is relative magnitude, never direction or guaranteed profit. A hot tile means "look here," not "buy this." A large modeled edge can reflect genuine mispricing *or* a market the model is reading wrong because of news it hasn't fully ingested. Read the reasoning.
  • •Categories cap at the top eight by volume. A niche category outside the top eight won't get its own tile in the default view — filter to it directly if it's your focus.
  • •Edge is a model estimate, not a fact. The pp figure is Predite's opinion of fair value versus market price. Confidence varies market to market; the Scanner exposes the per-market confidence the Heatmap aggregates away.

Plan requirements

The Market Heatmap is included on the Pro plan ($59/mo) and the Bot plan ($99/mo). It is not available on Starter ($29/mo) — Starter accounts will see an upgrade prompt at /dashboard/heatmap. Heatmap access travels with the same tier that unlocks the Realtime Stream and Arbitrage Detector, since all three rely on the higher-frequency market data feed.

Live order execution (placing real trades, TWAP orders, copy-trading, and automated bots) requires the Bot plan ($99/mo) regardless of Heatmap access. On Pro you can discover and analyze freely and validate ideas with paper trading; on Bot you can also act on what the Heatmap surfaces directly from inside Predite.

Related Docs

← All Documentation